A History of Diffusion Research 39
CHAPTER 2
A History of
Diffusion Research
Ironically, it almost seems as if diffusion research in the various research
traditions can be said to have been independently invented! Indeed, diffu-
sion researchers in the several traditions which we have examined scarcely
know of each other's existence.
Elihu Katz et al (1963),
"Traditions of Research on the
Diffusion of Innovations."
Diffusion research is thus emerging as a single, integrated body of con-
cepts and generalizations, even though the investigations are conducted
by researchers in several scientific disciplines.
Everett M. Rogers with F. Floyd
Shoemaker (1971, p. 47),
Communication of Innovations: A
Cross-Cultural Approach.
RESEARCH ON THE DIFFUSION OF INNOVATIONS started in a series
of independent intellectual enclaves during its first several decades.
Each of these disciplinary cliques of diffusion researchers studied one
kind of innovation; for example, rural sociologists investigated the
diffusion of agricultural innovations to farmers while educational
researchers studied the spread of new teaching ideas among school
personnel. Despite the distinctiveness of these approaches to diffusion
research, each "invisible college" uncovered remarkably similar find-
ings; for example, that the diffusion of an innovation followed an
s-shaped curve over time and that innovators had higher socioeco-
nomic status than later adopters.
My main motivation for writing my first book on this topic, Diffu-
sion of Innovations (Rogers, 1962), was to point out the lack of diffu-
38
sion of diffusion-research, and to argue for greater awareness among
the diffusion research traditions. A research tradition is a series of in-
vestigations on a similar topic in which successive studies are influ-
enced by preceding inquiries. Essentially, each research tradition is an
"invisible college" of researchers (or at least a "department" in such
an invisible college), a network of scholars who are spatially dispersed
but who are closely interconnected by exchanging research findings
and other scientific information.
But in the mid-1960s a gradual breakdown began to occur in the
formerly impermeable boundaries between the diffusion research tra-
ditions. Evidence"of this trend was provided by Rogers with Shoe-
maker (1971, pp. 46-47), who computed an index of cross-tradition
citations for each diffusion publication available by 1968; this index
was the number of research traditions (other than the author's) repre-
sented in the footnotes and bibliography of each empirical diffusion
publication. The average index score (per diffusion publication)
hovered at less than 1.0 during the 1940s, 1950s, and early 1960s. But
between 1965 and 1968, the average score suddenly doubled. Clearly
there was a trend toward breaking down the "paper curtains" among
the diffusion research traditions.
This trend toward a more unified and cross-disciplinary viewpoint
in diffusion research has continued until today; every diffusion
scholar is fully aware of the parallel methodologies and results in the
other traditions. All of the diffusion research traditions have now
merged, intellectually, toward one invisible college, although diffu-
sion studies are still conducted by scholars in several different
disciplines. But this merger of diffusion approaches has not been an
unmixed blessing. In fact, diffusion studies have begun to display a
kind of bland sameness, as they pursue a small number of research
issues with rather stereotyped approaches. It seems that the narrow
perspectives of diffusion scholars in an earlier era has been replaced
lately by an unnecessary and unhealthy standardization in diffusion
research. An observer might wonder whether the old days of separate
and varied research approaches might have been a richer intellectual
approach than the present era of well-informed sameness.
A major theme of this chapter is the story of the recent merging of
the diffusion research traditions, and both the good and the bad con-
sequences of this intellectual convergence. In this chapter, we will ad-
dress such questions as: where did diffusion research come from?
How and why did it grow to its present position of wide recognition by
scholars, and its widespread use and application by policy makers?
 Diffusion of Innovations
40
How has acceptance of the classical diffusion model limited the origi-
nality and appropriateness of the work of diffusion researchers?
We think this chapter is important because anyone who wishes to
learn about and understand diffusion research ought to know the his-
tory of how it attained its present status. I have been involved in diffu-
sion research since 1954, and so much of the twenty-eight-year history
described here is one that I know personally. I consider myself a loyal
and sympathetic, but critical, participant in the history of diffusion
research. That viewpoint should be kept in mind when reading the
pages that follow.
The Beginnings of Diffusion Research in Europe
Problems with the diffusion of innovations have been recognized for a
long time, and thus it should not be surprising that the roots of diffu-
sion research extend back to the European beginnings of social
science.
Gabriel Tarde and The Laws of Imitation
Gabriel Tarde, one of the forefathers of sociology and social psychol-
ogy, was a French judge around the turn of the century who kept an
analytical eye on trends in his society as represented by the legal cases
that came before his court. Tarde observed certain generalizations
about the diffusion of innovations that he called "the laws of imita-
tion," and this became the title of his influential book, which was
published in 1903. The purpose of his scholarly observations, Tarde
(1903, p. 140) said, was 'to learn why, given one hundred different in-
novations conceived of at the same time—innovations in the form of
words, in "mythological ideas, in industrial processes, etc.—ten will
spread abroad while ninety will be forgotten."
Gabriel Tarde was undoubtedly an intellectual far ahead of his
time in his thinking about diffusion. Even though he used slightly dif-
ferent concepts from those employed in the present book (for exam-
ple, what Tarde called "imitation" is today called the "adoption" of
an innovation), this sociological pioneer was on to several of the main
research issues that were to be pursued by diffusion scholars in later
decades, using more quantitative approaches. For example, as the
A History of Diffusion Research 41
above quotation indicates, Tarde identified the adoption or rejection
of innovations as a crucial research question. He observed that the
rate of adoption of a new idea usually followed an s-shaped curve over
time. At first, only a few individuals adopt a new idea, then the rate of
adoption spurts as a large number of individuals accept the innova-
tion, and finally the adoption rate slackens. Very astutely, Tarde rec-
ognized that the "take off" in the s-curve of adoption began to occur
when the opinion leaders in a system used the new idea. So diffusion
network thinking was involved in Tarde's explanation of the s-curve,
even though he did not use such present-day concepts as networks,
homophily, and heterophily. For example, Tarde (1969, pp. 29-30)
observed that an innovation is first adopted by an individual who is
socially closest to the source of the new idea, and that it then spreads
gradually from higher-status to lower-status individuals. Further,
Tarde (1969, p. 27) proposed as one of his most fundamental "laws of
imitation" that the more similar an innovation is to those ideas that
have already been accepted, the more likely the innovation is to be
adopted (we discussed in Chapter 1 that the perceived compatibility of
an innovation is related to its rapid rate of adoption).
To Gabriel Tarde, the diffusion of innovations was a basic and
fundamental explanation of human behavior change: "Invention and
imitation are, as we know, the elementary social acts" (Tarde, 1969,
p. 178). So Tarde was one of the European grandfathers of the diffu-
sion field. But his creative insights were not followed up immediately
by empirical studies of diffusion. That was not to happen until after a
lapse of almost forty years. Perhaps social scientists of Tarde's day
lacked the methodological tools to mount diffusion studies; maybe
they were just not inclined to follow up on his leads. In any event, his
suggested approach to diffusion research lay fallow for several dec-
ades, until an invisible college of American scholars was to coalesce
around Tarde's "laws of imitation." — "
The British and German-Austrian Diffusionists
Another root in the ancestry of diffusion research was a group of early
anthropologists that evolved in England and in Germany-Austria
soon after the time of Gabriel Tarde in France (although they were not
influenced by his writings). These anthropologists are called the
"British diffusionists" and the "German-Austrian diffusionists."
The viewpoint of each group was similar. Diffusionism was the point
 Diffusion of Innovations
42
of view in anthropology that explained change in a given society as a
result of the introduction of innovations from another society. The
diffusionists claimed that all innovations spread from one original
source, which, of course, argued against the existence of parallel in-
vention (today we know that such parallel invention of new ideas has
frequently occurred in history).
The diffusionism viewpoint does not have much of a following to-
day, owing to the extreme claim of the diffusionists that all social
change could be explained by diffusion alone. The dominant view-
point now is that social change is caused by both invention (the proc-
ess by which a new idea is discovered or created) and diffusion, which
usually occur sequentially. Viewed in retrospect, we see that the main
contribution of the European diffusionists was in their calling the im-
portance of diffusion to the attention of other social scientists
(Kroeber, 1937, pp. 137-142). The diffusionists would have had
greater impact if they had not so overstated their case.
The scholars who picked up on the work of the European diffu-
sionists most directly, as one might expect, were anthropologists,
especially those in the United States who, beginning in the 1920s,
began to investigate the diffusion of innovations.
The Rise of Diffusion Research Traditions
The anthropological diffusion researchers constitute the oldest of the
nine diffusion research traditions.* In the following sections of this
chapter we shall trace the intellectual ancestry of each of these nine
traditions, as they help us to understand the history of diffusion
research. Each research tradition consists of an academic discipline
(for example, anthropology, marketing, geography) or a subdiscipline
(for instance, early sociology, rural sociology, medical sociology).
Each tradition usually has concentrated on investigating the diffusion
of one main type of innovation: for example, rural sociologists have
specialized in farm innovations. Table 2-1 shows, for each tradition,
* The exact number of major diffusion research traditions is, of course, somewhat ar-
bitrary. We chose these nine because they represent the relatively greatest number of
empirical diffusion publications (an exception is the early sociology tradition, which
is included because of its considerable influence on most of the other traditions which
develop later). The nine traditions represent a total of 2,585 of the 3,085 publications
available in late 1981, or 84 percent of the total diffusion reports then available.
A History of Diffusion Research 43
the main types of innovations studied, methods of data gathering and
analysis, the unit of analysis, and the types of findings. This overview
and comparison of the nine diffusion research traditions is com-
plemented by the following narrative description of each tradition.
Paradigms and Invisible Colleges
Any given field of scientific research begins with a major break-
through or reconceptualization, called a "revolutionary paradigm"
by Kuhn (1970), that provides a new way of looking at some phenome-
non. A paradigm is a scientific approach to some phenomena that
provides model problems and solutions to a community of scholars.
Recognition of a new paradigm typically sets off a furious amount of
intellectual effort as promising young scientists are attracted to the
field, either to advance the new conceptualization with their research
or to disprove certain of its aspects. Gradually, a scientific consensus
about the field is developed, and, perhaps after several generations of
academic scholars, the invisible college (an informal network of
researchers who form around an intellectual paradigm to study a com-
mon topic) declines in scientific interest as fewer findings of an ex-
citing nature are turned up. (We show the invisible college of rural
sociologists studying diffusion later in this chapter, in Figure 2-2.)
These are the usual stages in the normal growth of science, Kuhn
(1970) claims. The research process is a very social activity in which
crucial decisions are influenced by a network of scientists, who are
organized around one important research idea.
As Abraham Kaplan (1964, p. 28) stated in his important book on
research methods, "Give a small boy a hammer, and he will find
everything he encounters needs pounding." Most scientists in any
field are much like Kaplan's little boy; in deciding which research
problem to study and exactly how to study it, they face uncertainty. In
behavior parallel to that of the potential adopters of an innovation,
scientists rely on the subjective experiences of their peers. An invisible
college centered in an intellectual paradigm provides the typical scien-
tist with the information he or she needs to reduce the uncertainty of
the research process. Of the many alternative directions that a re-
search project might pursue, a paradigm structures a researcher
toward one general approach. Thus, the paradigm and the invisible
college of scientists that follow the paradigm provide a researcher with
a source of security and stability in the uncertain world of research.
ft
46 Diffusion of Innovations

Research on the diffusion of innovations followed these rise-and-

fall stages rather closely (Crane, 1972), although the final stage of
demise does not seem to have begun. The hybrid corn diffusion study
by Bryce Ryan and Neal Gross (1943), described in Chapter 1, set
forth a new approach to the study of diffusion,* that was soon fol-
lowed up by an increasing number of scholars. Within ten years (by
1953), over 146 diffusion researches were completed; during the next
decade (by 1963), another 647; and by 1973, another 1,417. In 1981
there were over 3,085 publications about the diffusion of innovations,
including about 2,297 empirical research reports and 788 other writ-
ings (Figure 2-1). Thus, we see that the amount of scientific activity
involved in investigating the diffusion of innovations has increased at
a very sharp rate since the revolutionary paradigm appeared about
forty years ago, as Kuhn's (1970) theory of the growth of science
would predict.
Diffusion research is a particular type of communication research
(as we explained in Chapter 1), but it began outside of the academic
field of communication. Mostly this was a matter of timing, as for ex-
ample, the Ryan and Gross (1943) hybrid corn study preceded the first
university centers or departments of communication by about ten
years. The diffusion research approach was taken up in a variety of
fields: education, anthropology, medical sociology, marketing,
geography, and in rural sociology. Each of these disciplines pursued
diffusion research in its own specialized way, and for some time
without much interchange with the other diffusion research tradi-
tions, at least until the early 1960s when the boundaries between the
traditions began to break down.
But before we describe this intellectual integration in the 1960s, we
must return to the beginnings of the anthropological research tradi-
tion on diffusion, in the 1920s.

The Anthropology Research Tradition

The anthropology tradition is not only the oldest of the nine traditions
analyzed in this book, it is also the most distinctive in its methodolog-
* Perhaps some question might be raised as to whether formulation of the diffusion
approach truly constituted a paradigm, or only a "quasi-paradigm." The diffusion
conceptualization was distinctive in the social sciences at the time of its formulation in
the 1940s, and it certainly set off a great number of following researches. So it seems to
meet our definition of a paradigm. But certainly the diffusion paradigm was not as
revolutionary as Copernicus' astronomy, Newtonian physics, Darwin's evolution, or
Einstein's theory of relativity. In comparison with these great ideas of science, the dif-
fusion model is but a mini-paradigm.
 Diffusion of Innovations
48
ical approach to studying diffusion. As a rule, most anthropologists
are taught by their mentors to distrust numbers as a basis for describ-
ing and analyzing social behavior. The quantitative approach, based
as it necessarily is on a social scientist learning'' a little about a lot," is
anathema to the anthropologist who strongly prefers to learn "a lot
about a little." This point of view means that the anthropologist who
studies diffusion usually prefers to avoid using such tools as personal
interviews, random sample surveys, and computer data-analysis. In
fact, such "number crunching" with a computer is abhorrent to most
anthropologists.
Anthropologists prefer to gather diffusion data more directly
from their respondents, by means of participant observation, a com-
mitment by a researcher to adopt the perspective of the respondents by
sharing their day-to-day experiences. In order to get into his or her
subjects' skins, an anthropologist often goes to live among them,
seeking to empathize into their everyday roles. Obviously, such a
total-immersion approach requires a great deal of patience on the part
of the anthropologist field researcher, who may have to wait for a long
time indeed for what he or she has come to observe (such as diffusion
and adoption behavior) to occur. The participant-observation method
not only requires a long time for data gathering (anthropologists often
live among their respondents for several years), it also means that an-
thropologists are limited to studying diffusion in rather tiny systems,
often a single village. Most anthropological research is a one-person
operation, and the investigator is therefore limited to what he or she
can observe in a limited setting. The results of such inquiry provide
valuable insights into the microscopic details of diffusion and adop-
tion. But one cannot be very certain that the results of anthropological
diffusion studies are generalizable. For instance, to what extent can
the administrators of the public health service in Peru apply the find-
ings from Wellin's (1955) anthropological investigation of the failure
of the water-boiling campaign in Los Molinos (described in Chapter 1):
to other Peruvian villages? Does Los Molinos have special characteris-
tics that affected the adoption and rejection of water boiling? Would
similar diffusion problems occur in other Peruvian villages? We do
not know.
There are, however, several special advantages of anthropological
research on diffusion. For one thing, if the anthropologist is suc-
cessful in his or her attempt to empathize with the respondents of
study, the ensuing account of diffusion will tell the story from the
respondents' viewpoint, conveying their perceptions of the innovation
A History of Diffusion Research 49
and of the change agency with an in-depth understanding that other
social scientists can seldom match. This perspective helps the anthro-
pologist overcome the pro-innovation bias of most other diffusion re-
searchers. If anything, anthropologists sometimes seem to display an
inclination toward anti-innovation. Through total immersion in the
respondents' system, the anthropologist gains a holistic perspective of
the lifestyles, world views, and social relationships of the respondents.
This capacity of anthropologists to understand the total culture of
their individuals of study, coupled with their over-time data gather-
ing, provides the anthropological diffusion scholars with a unique
means of understanding the consequences of innovation. It is no acci-
dent that much of the research featured in our Chapter 11 was carried
out by anthropologists.
In addition to their useful contributions to our understanding of
consequences, a good deal of anthropological research also has been
conducted on the relationship of an innovation's compatibility with
cultural values, to the innovation's rate of adoption.* In many of their
research accounts, anthropologists show that the planners and offi-
cials in charge of development programs failed to account fully for the
cultural values of the expected adopters of an innovation. As a result,
the diffusion program often failed, or at least it led to unanticipated
consequences.
Compared to other research traditions, anthropology has been
more concerned with the transfer of technological innovations from
one society to another (as compared to the diffusion of a new idea
within a society or system). This emphasis on cross-cultural diffusion
is consistent with anthropologists' interest in the concept of culture,
one of their favorite intellectual tools. An early illustration of this type
of investigation was Wissler's (1923, pp. 111-121) study of the diffu-
sion of horses from Spanish explorers to American Indian tribes.
More contemporary studies of cross-cultural diffusion in the anthro-
pological diffusion tradition comprise research that evaluates the ef-
fectiveness of development programs in which Western technologies
are introduced in the developing countries of Latin America, Africa,
and Asia.
In part owing to their early appearance on the diffusion research
scene, anthropologists have influenced the other eight diffusion
Summaries of anthropological evidence on this point are provided by Spicer (1952),
Barnett (1953), Arensberg and Niehoff (1964), and Niehoff (1966). Linton (1936,
PP. 324-336) was one of the first scholars to recognize the relationship of the per-
ceived characteristics of an innovation to its rate of adoption.
 Diffusion of Innovations
50
research traditions, particularly early sociology and rural sociology.
The other traditions have seldom used participant observation as their
data-gathering methodology, but they have carried forward into
quantitative research certain of the theoretical leads pioneered by an-
thropology diffusion scholars.
Early Sociology
The intellectual tradition that we refer to as "early sociology" traces
its ancestry to the French sociologist Gabriel Tarde, but most of the
research publications in this tradition appeared from the late 1920s to
the early 1940s (about the same time that the anthropology diffusion
tradition was getting under way). The social significance of the early
sociology tradition lies neither in its volume of investigations (there
are only ten) nor in the sophistication of its research methods but in
the considerable influence of early sociologists upon later diffusion
researchers.
Most early sociologists traced the diffusion of a single innovation
over a geographical area like a state or a region. The motivating in-
terest of the early sociologists was primarily in the diffusion of innova-
tions that contributed to social change. With the exception of Bowers
(1937; 1938), who investigated the diffusion of ham radio sets, early
sociologists did not emphasize the innovation-decision process nor did
they concentrate upon the process by which opinion leaders influ-
enced others in their system to adopt or reject a new idea.
Bowers' (1937; 1938) investigation was probably the first study in
the early sociology tradition that used primary data from respondents,
in addition to data from secondary sources like government records.
He contacted a sample of 312 ham-radio operators in the United
States by mailed questionnaire in order to determine the influences
that led to their adoption of the radios. Bowers (1938) was the first
researcher to find that interpersonal channels are more important
than mass-media channels for later adopters than for earlier adopters.
The number of amateur radio operators in the United States had in-
creased sharply from about 3,000 in 1914 to 46,000 in 1935, and
Bowers determined that this adopter distribution followed an
s-shaped normal curve when the number of adopters were plotted by
year. Bowers also related such ecological factors as city size and region
in the United States to the rate of adoption of ham radios. Like others
in the early sociology tradition, Bowers thus correlated ecological fac-
tors to innovativeness.
A History of Diffusion Research 51
The ten studies in the early sociology diffusion tradition differed
from their anthropological counterparts in that they used quantitative
data analysis, a methodological approach that was to be followed by
other research traditions. But the intellectual paradigm that was to set
off widespread research on the diffusion of innovations had not yet
happened. Creation of this paradigm had to wait for the rural sociol-
ogy tradition.
Rural Sociology
The research tradition that can claim major credit for initially forming
the intellectual paradigm for diffusion research, and that has pro-
duced the largest number of diffusion studies and over the longest
period of years, is rural sociology. Dominance of the diffusion field by
rural sociology, however, as indexed by the percentage of all diffusion
studies that were completed by rural sociologists, has declined over
the past twenty years as other diffusion research traditions have grown
more rapidly in size. Up to 1964, 423 of the 950 diffusion publications
(45 percent) were in the rural sociology tradition. From 1965 to 1969,
only 225 (26 percent) of the 849 diffusion publications were in rural
sociology, and this percentage dropped further, to 14 percent (100 of
708 diffusion publications) from 1970 to 1974. Finally, since 1974,
only 8 percent (45 of 578 publications) of all diffusion studies are in
rural sociology. Today the rural sociology tradition no longer hogs the
field of diffusion research; this tradition has faded in both its relative
and absolute role in the diffusion field. But 791 of the 3,085 diffusion
publications available in 1981 (26 percent) were by rural sociologists.
Table 2-2 shows that rural sociology is still number one in its share of
all diffusion studies. Perhaps it is a healthy sign that diffusion
research is becoming more multidisciplinary.
Rural sociology is a subfield of sociology that focuses on the social
problems of rural life. Most rural sociologists are employed by land-
grant universities in colleges of agriculture. These agriculture schools
have three main functions: (1) to teach students, (2) to conduct
research on agricultural problems, so as to help farmers and agricul-
tural businesses, and (3) to operate a state extension service to diffuse
the agricultural innovations (coming from research) to potential
adopters, mainly farmers. The state colleges of agriculture and their
research and extension subunits, the state agricultural experiment sta-
tions, and the state agricultural extension services are dominated by
administrators and scientists specializing in agricultural production
 Diffusion of Innovations
54
fields (for example, crop growing, milk production, beef farming, and
horticultural production). In such an organization, where the main
value is on raising farm production, most of the activities of rural
sociologists are considered rather superfluous by the agricultural
scientists who run the state colleges of agriculture.
Except for diffusion research. It can provide helpful leads to agri-
cultural researchers about how to get their scientific results put into
use by farmers. And diffusion research is greatly appreciated by exten-
sion service workers, who depend on the diffusion model as the main
theory guiding their efforts to transfer new agricultural technologies
to farmers (Rogers et al, 1982a). So diffusion research fits well with
the heavy value on agricultural production that dominates colleges of
agriculture. Other research by rural sociologists on such important
social problems as the increase in rural crime rates, the decrease in
farm population through migration to cities, and rural health prob-
lems, is not so well appreciated by the agricultural biologists who run
state colleges of agriculture. Under these organizational conditions, it
is not surprising that the diffusion of agricultural innovations became
a popular research topic among rural sociologists.
THE HYBRID CORN STUDY
Although a couple of diffusion studies had been completed during the
1920s and 1930s, the Ryan and Gross (1943) investigation of the diffu-
sion of hybrid-seed corn, more than any other study, influenced the
methodology, theoretical framework, and interpretations of later
students in the rural sociology tradition, and in other research tradi-
tions as well. Dr. Bryce Ryan was a professor of rural sociology at
Iowa State University, the state land-grant school at Ames. In 1941,
he convinced the Iowa Agricultural Experiment Station (the research
branch of the college of agriculture) to fund his proposed investiga-
tion of the spread of hybrid seed to Iowa farmers. This innovation was
something of a success story for Iowa State University. The develop-
ment of hybrid seed corn had resulted from twenty years of genetic
research by agricultural scientists at Ames; finally, in 1928 hybrid seed
was made available to Iowa farmers, promoted by the Iowa Agricul-
tural Extension Service and by the commercial seed companies that
marketed the seed. As was pointed out in Chapter 1, the hybrid vigor
of the new seed typically increased corn yields on Iowa farms by about
20 percent, hybrid corn varieties withstood drought better than the
A History of Diffusion Research 55
open-pollinated seed they replaced, and hybrid corn was better suited
to harvesting by mechanical corn pickers. Corn was the main farm
crop in Iowa in the 1930s; in fact, Iowa's official state song bills it as
"the tall corn state." It is no surprise that under these conditions, the
hybrid seed was adopted rapidly. By 1941, about thirteen years after
its first release, the innovation was adopted by almost 100 percent of
Iowa farmers.
Presumably, administrators in the Iowa Agricultural Experiment
Station sponsored Professor Ryan's diffusion study because they
wanted to improve their understanding of this case of successful diffu-
sion in order to learn lessons that might be applied to the diffusion of
future farm innovations. These officials may also have been puzzled
and frustrated as to why such an obviously advantageous innovation
as hybrid corn was not adopted more rapidly. They wondered, for ex-
ample, why some farmers waited thirteen years to adopt, a period dur-
ing which they were surrounded by neighbors who were using the in-
novation successfully.
In the summer of 1941, Neal Gross, a new graduate student in
rural sociology, was hired by Professor Ryan as a research assistant on
the hybrid corn diffusion project. They selected two small Iowa com-
munities, located to the west of Ames, and proceeded to interview per-
sonally all of the 259 farmers living in the two systems. Using a struc-
tured questionnaire, Neal Gross, who did most of the data gathering,
interviewed each respondent about when he decided to adopt hybrid
corn (the year of adoption was to become the main dependent variable
in Ryan and Gross' data analysis), the communication channels he
had used at each stage in the innovation-decision process, and how
much of his corn acreage he had planted in hybrid (rather than open-
pollinated) seed each year. In addition to these recall data about the
innovation, the rural sociologists also asked each respondent about
his formal education, age, farm size, income, travel to Des Moines
and other cities, readership of farm magazines, and other variables
that were later to be correlated with innovativeness (measured as the
year in which each farmer decided to adopt hybrid corn).
When all the data were gathered, Ryan and Gross converted the
farmers' interview responses into coded form (that is, to numbers).
These diffusion researchers analyzed the data by hand tabulation and
with a desk calculator (the use of a computer for data analysis did not
begin until about fifteen years later). Within a year, Neal Gross (1942)
completed his Master's thesis on the diffusion of hybrid corn, and
shortly thereafter Ryan and Gross (1943) published an article out of
56 Diffusion of Innovations
the study in the journal, Rural Sociology (this article is the most
widely cited publication out of the study, although there are several
others).
We described the main findings from the hybrid corn study in
Chapter 1, and there is no need to repeat them here. This classic diffu-
sion study headed later diffusion scholars toward pursuing certain
research questions such as: what variables are related to innovative-
ness? What is the rate of adoption of an innovation, and what factors
(like the perceived attributes of the innovation) explain this rate?
What role do different communication channels play at various stages
in the innovation-decision process? These research directions have
continued to dominate almost all diffusion research since 1943. The
intellectual influence of the hybrid corn study reached far beyond
Iowa, the study of agricultural innovations, and even outside the rural
sociology tradition of diffusion research. The research paradigm cre-
ated by the Ryan and Gross investigation became the academic tem-
plate that was to be mimicked, first by other rural sociologists in their
agricultural diffusion researches, and then by almost all other diffu-
sion research traditions (whether they knew it or not).
The Iowa hybrid corn study has left an indelible stamp on the field
of diffusion research up to the present. This lasting influence is not
completely beneficial, intellectually speaking. An overly close copying
of the classical diffusion paradigm by later researchers, who were
often investigating diffusion of a quite different type, led to inap-
propriate methodologies and mistaken theoretical thrusts. Criticisms
such as these, caused by the dominance of the classical paradigm, are
discussed in Chapter 3. We argue that the overwhelming relative ad-
vantage of hybrid corn (over open-pollinated seed) may have con-
tributed to both the pro-innovation bias of later diffusion studies and
to the lack of research attention paid to the consequences of techno-
logical innovations. Because the effects of hybrid corn were so ob-
viously beneficial, it was easy to assume that the consequences of
other innovations would also be positive.
In addition to structuring the diffusion paradigm theoretically, the
Ryan and Gross hybrid corn study also established a prototypical
methodology for going about a diffusion investigation: one-shot sur-
vey interviews with the adopters of an innovation, who were asked to
recall their behavior and decisions regarding the innovation. Thus, the
typical research design for studying diffusion was established in 1941.
It has lived on, with only rare and minor modifications, to the present
day. The alternate methodological paths that were not taken by diffu-
sion scholars represent a shortcoming in the field today.
A History of Diffusion Research 57
THE INVISIBLE COLLEGE OF RURAL SOCIOLOGY DIFFUSION
RESEARCHERS Is FORMED
During the 1950s, an explosion occurred in the number of diffusion
studies by rural sociologists. Important pioneers in this tradition were
(1) Dr. Eugene A. Wilkening, who moved to the University of Wis-
consin in the early 1950s after several years of excellent diffusion
studies in North Carolina, and (2) Dr. Herbert F. Lionberger of the
University of Missouri. A third center of research and training in
agricultural diffusion was at Iowa State University where Professors
George M. Beal and Joe M. Bohlen carried forward the diffusion
studies launched by Ryan and Gross. New Ph.D.s in rural sociology,
produced at Madison, Columbia, and Ames in the 1950s, then became
professors of rural sociology at other state land-grant universities
where they, in turn, established diffusion research programs. In fact, I
was one of these diffusion research missionaries.
Crane (1972, p. 188) studied the invisible college of diffusion
researchers in the rural sociology tradition in the mid-1960s, and con-
cluded that it was a highly interconnected network of scholars who
shared a common theoretical-methodological framework. Dominat-
ing the network were two large cliques,* one composed of twenty-
seven scholars and the other of thirty-two researchers; each centered
in a leading scholar of diffusion whose network links reached out to
former Ph.D. students and to the students of those students. Smaller
cliques of thirteen, twelve, seven, etc. scholars were highly connected
to the two major cliques (Figure 2-2). This communication structure
of the network of rural sociology diffusion researchers provided con-
sensus and coherence to the field; it meant that these scholars shared a
common framework in studying diffusion, and that they were kept
abreast of each others' research findings. This helped the field to
progress in an ordered direction toward its research goals. There was a
cumulative nature to these research directions, as each study built
upon the accomplishments of previous work. Unfortunately, it also
meant that radical deviations from the diffusion paradigm were im-
plicitly discouraged or stultified. Some of the reality of diffusion was
ignored, because it was not part of the accepted diffusion paradigm.
Another key factor in the 1950s-1960s proliferation of the rural
sociology diffusion research tradition, in addition to the intercon-
nectedness of the invisible college of scholars, was the availability of
A clique is a subsystem whose elements interact with each other relatively more fre-
quently than with other members of the communication system.

Figure 2-2. Communication network structure of rural sociologists study-
ing diffusion as of 1967, on the basis of their collaboration.
The invisible college of rural sociologists studying diffusion was rela-
tively interconnected in 1967 when Crane gathered these network data by
mailed questionnaire from the 221 scholars in this diffusion research tradi-
tion. The cliques shown here were identified by a network-analysis computer
program so that each clique included individuals who interact more fre-
quently with each other than with others. For the sake of simplicity, we have
not shown the links within each clique, nor have we shown isolates in the
sociogram. Direct collaboration between individuals in a pair of cliques is
shown as a solid line, and a broken line indicates indirect collaboration (indi-
cating that any member of a clique is linked to an individual in another clique
through someone else). The two largest cliques, containing twenty-seven and
thirty-two researchers, respectively, provide connectedness to the entire in-
visible college; if they were removed, the network would tend to decompose.
The four largest cliques include all eight of the "high producers" (each of
whom had ten or more diffusion publications); most of the clique members
were their collaborators or students. All eight high producers were in com-
munication with one another about current research. As in other invisible
colleges that have been studied, the most productive scientists are leaders of
cliques, and their contacts with each other link the cliques into a network.
However, 101 of the 221 researchers are isolates or members of cliques not
connected to the rest of the network. Few of these 101 individuals are pro-
ductive scholars; many just completed their Master's or Ph.D. theses but
had no further publications. Some do not live in the United States, having
returned to their home country after finishing graduate study in America.
Source: Based on data reported by Crane (1972, p. 188).
A History of Diffusion Research 59
research funds. During this period, the state agricultural experiment
stations, together with the U.S. Department of Agriculture which par-
tially funds the state agricultural research, were producing a tremen-
dous outpouring of farm innovations: weed sprays, chemical fer-
tilizers, new crop varieties, chemical feeds for livestock, and new farm
machinery. The result was an "agricultural revolution" in which the
number of persons fed and clothed by the average American farmer
shot up from fourteen in 1950, to twenty-six in 1960, to forty-seven in
1970. This rapid increase in agricultural productivity rested not only
on the availability of farm innovations, but also on their effective dif-
fusion to American farmers.
That is where the rural sociologists came in. Their diffusion studies
helped show agricultural extension workers how to communicate
these new technological ideas to farmers, and thus how to speed up the
diffusion process. Thanks to Ryan and Gross (1943), the rural sociolo-
gists had an appropriate paradigm to guide their diffusion studies.
Thanks to the agricultural revolution of the 1950s, these diffusion
scholars were in the right place (state university colleges of agriculture)
at the right time. The result was a proliferation of diffusion studies by
the rural sociology tradition: 185 by 1960, 648 by 1970, and 791 by
1981.
RURAL SOCIOLOGY DIFFUSION RESEARCH IN DEVELOPING NATIONS
By the 1960s, American rural sociologists went international. This
decade marked a large-scale attempt to export the land-grant univer-
sity/agricultural extension service complex to the developing nations
of Latin America, Africa, and Asia. With funding from the U.S.
Agency for International Development (AID) and from private foun-
dations, U.S. land-grant universities created overseas campuses in
which American faculty members taught, conducted agricultural re-
search, and advised extension services and other development pro-
grams (Rogers et al, 1982a). Rural sociologists were part of this over-
seas operation, and they (in collaboration with graduate students
from these developing nations that they had trained) launched diffu-
sion studies in peasant villages. Agricultural development was the
main thrust of these international activities, so it was natural that the
topic of diffusion of farm innovations should be pursued. In addition,
rural sociologists branched out to investigate the diffusion of nutri-
hon, health, and family-planning innovations to villagers.
60 Diffusion of Innovations
The early 1960s marked the beginning of a sharp take off in the
number of diffusion studies in developing countries (Figure 2-1). Pio-
neering ventures in this direction by Syed A. Rahim (1961) in Bangla-
desh, and by Paul J. Deutschmann and Orlando Fals Borda (1962b) in
Colombia, suggested that new ideas spread among peasants in villages
in a generally similar pattern to their diffusion in more media-
saturated settings like the United States and Europe. The diffusion
process, and the concepts and models used to analyze it, seemed to be
cross-culturally valid, at least in the sense that comparable results were
found in the new settings. In later years, however, the applicability of
the diffusion paradigm that was exported from the United States to
developing nations, began to be questioned.
There were compelling reasons for the fast growth of diffusion
studies in developing countries in the 1960s. Technology was assumed
to be at the heart of development, at least as development was concep-
tualized at that time, so microlevel investigations of the diffusion of
technological innovations among villagers were of direct relevance to
development planners and other government officials in developing
nations. These research results, and the general framework of diffu-
sion, provided both a kind of theoretical approach to planning devel-
opment programs and an evaluation procedure for measuring the suc-
cess of development activities.
The number of diffusion researches in developing nations totaled
only about 71 by 1960 (14 percent of all diffusion studies), but rose
steeply to 601 in 1970 (31 percent of the total), and to 912 by 1981,
when 30 percent of all diffusion studies had been conducted in Latin
America, Africa, and Asia. By no means were all of these studies con-
ducted by rural sociologists, but this tradition played a pioneering role
in beginning diffusion research in developing nations (although since
the mid-1970s, rural sociologists were conducting only a few diffusion
studies in developing nations). In the 1970s, there began to appear
criticisms of the diffusion paradigm as it was applied in Latin
America, Africa, and Asia (which will be discussed in Chapter 3).
HARD TOMATOES AND HARD TIMES
During the 1970s some American rural sociologists began to question
whether conducting research on the diffusion of agricultural innova-
tions was indeed their most useful role, as social scientists of rural
society. Such a questioning attitude was given a big boost by a radi-
A History of Diffusion Research 61
cally critical book written by James Hightower (1972), Hard Toma-
toes, Hard Times: The Failure of America's Land-Grant College
Complex. The author used as a spectacular illustration the case of
mechanized tomato harvesting, which required that farmers plant to-
mato varieties that are still very firm when they ripen. Both the har-
vesting machine and the hard tomato varieties were developed by agri-
cultural researchers at state colleges of agriculture. One benefit of
these innovations was cheaper tomato prices for the consumer, but
unfortunately many consumers did not like the hard tomatoes. They
expected ripe tomatoes to be soft. Indeed, the hard tomatoes con-
tained somewhat fewer vitamins than the older, soft varieties. Fur-
ther, the mechanized tomato harvesters put thousands of farm labor-
ers out of work, and drove thousands of small farmers, who could not
afford to buy the expensive harvesting machines, out of tomato pro-
duction (the consequences of the tomato harvester are described in
more detail in Chapter 4).
Hightower (1972) claimed that the state colleges of agriculture
were responsible for the "agricultural revolution" in the United States
through their development and diffusion of farm innovations, but
that they had almost totally ignored the consequences of these techno-
logical innovations. Hightower said this technological irresponsibility
amounted to a failure on the part of U.S. colleges of agriculture. This
critical analyst showed that almost all of the professional resources of
the publicly supported land-grant colleges went into (1) biological
science, to develop innovations, and (2) agricultural extension service
activities to diffuse these new ideas to farmers. This overemphasis on
agricultural production technology meant that social science research
on the consequences of innovation was severely shortchanged. High-
tower's criticisms hit rural sociologists especially hard; they had been
investigating diffusion for the past twenty years or so, in order to
speed up the rate of adoption, instead of studying the consequences of
technology and what could be done about the social problems stem-
ming from the agricultural revolution in the United States.
While a number of useful diffusion studies continue to be con-
ducted in the rural sociology tradition today, more attention is now
being paid to investigating the consequences of agricultural technol-
ogy (several of these researches will be discussed in Chapter 11). Fur-
ther, some rural sociologists have become much more questioning of
the emphasis placed upon agriculture production technology by col-
leges of agriculture. If the result is increased agricultural production in
the United States at the cost of driving many farm families out of
62
Diffusion of Innovations
agriculture, some rural sociologists wonder if colleges of agriculture
are really serving the U.S. farmer. Some rural sociologists have
become a kind of social conscience for U.S. colleges of agriculture. It
is a quite different role from that played by the rural sociology tradi-
tion prior to about 1970. And this is one reason why the interest of
rural sociologists in diffusion research has faded somewhat in recent
years.
Education
Although it is an important diffusion research tradition in terms of the
number of studies completed,* education is less important in terms of
its contribution to the theoretical understanding of the diffusion of in-
novations. But there is an exciting potential contribution to be made
by the education research tradition, stemming from the fact that
organizations are involved, in one way or another, in the adoption of
an educational innovation. Unlike U.S. farmers, who mainly make
optional innovation-decisions, most teachers and school administra-
tors are involved in collective and/or authority innovation-decisions.
Teachers, unlike farmers, work in organizations.
THE TEACHERS COLLEGE STUDIES
A majority of early educational diffusion studies were completed at
one institution, Columbia University's Teachers College, and under
the direction of one man, Dr. Paul Mort. This tradition traces its roots
to research in the 1920s and 1930s by Mort and others on local control
over school financial decisions (as opposed to federal or state influ-
ence on these decisions), and on whether this local control led to
school innovativeness. In short, the Columbia University education
diffusion studies set out to show that local school control was related
to innovativeness, which was thought to be a desirable characteristic
of schools.
The data in these studies were most often gathered by question-
* Education diffusion publications numbered 23 in 1961 (5 percent of all diffusion
work), 71 in 1968 (6 percent), and 336 in 1981 (11 percent of all diffusion
publications). Education ranks third among the nine diffusion traditions in terms of
number of publications (see Table 2-2).
A History of Diffusion Research 63
naires mailed to school superintendents or principals.* The unit of
analysis was the school system in almost all these investigations. The
Columbia University diffusion studies found that the best single
predictor of school innovativeness was educational cost per pupil. The
wealth factor almost appeared to be a necessary prerequisite for inno-
vativeness among public schools. One's stereotype of the rich sub-
urban school in the United States as highly innovative, was largely
confirmed by the early Teachers College studies. Further, Dr. Mort
and his fellow researchers found that a considerable time lag was re-
quired for the widespread adoption of new educational ideas: "The
average American school lags 25 years behind the best practice"
(Mort, 1953, pp. 199-200).
There is, of course, a wide range in the rate of adoption of educa-
tional innovations. For instance, it took kindergartens about fifty
years (from 1900 to 1950) to reach complete adoption by U.S. schools
(Mort, 1953). But driver training needed only eighteen years (from
1935 to 1953) to reach widespread adoption (Allen, 1956), and
modern math took only the five years from 1958 to 1963 (Carlson,
1965). Driver training and modern math were heavily promoted by
change agencies: insurance companies and auto manufacturers in the
case of driver training, and the National Science Foundation and
the U.S. Department of Education in the case of modern math. The
post-1958 aftermath of Sputnik caused public dissatisfaction with
U.S. education and marked the beginning of an active federal govern-
ment role in diffusing educational innovations. This recent involve-
ment by federal and state-level governments in educational diffusion
has somewhat eroded the degree of local school control which Mort
had originally set out to show was so valuable.
LATER STUDIES ON EDUCATIONAL DIFFUSION
After Paul Mort's death in 1959, Teachers College of Columbia
University lost its monopolistic control on educational diffusion.
More recent studies focused (1) upon teachers as respondents, rather
64 Diffusion of Innovations
than simply on school administrators, (2) on within-school as well as
school-to-school diffusion, and (3) on educational diffusion in devel-
oping nations. Many studies in the education tradition are sponsored
by the U.S. Department of Education, as a means to evaluate the
various diffusion programs that this government agency carries out.
Many other diffusion studies are conducted by graduate students in
education for their doctoral dissertations.
Two of the academic leaders in educational diffusion research are
Dr. Ronald G. Havelock of American University and Dr. Matthew B.
Miles of the Policy Research Institute in New York City; both have
been engaged in diffusion research for over fifteen years, and each has
written or edited a much-cited book (Miles, 1964; Havelock, 1969).
THE DIFFUSION OF MODERN MATH IN ALLEGHENY COUNTY
Probably the best piece of educational diffusion research is Dr.
Richard O. Carlson's (1965) analysis of the spread of modern math
among school administrators in Pennsylvania and West Virginia. He
studied the opinion leadership patterns in the diffusion networks for
modern math among school superintendents, variables correlated to
innovativeness, perceived characteristics of innovations and their rate
of adoption, and the consequences of one educational innovation:
programmed instruction.
But Carlson's study is most impressive in the insight that it pro-
vides into the diffusion networks through which modern math spread
from school to school in Allegheny County, Pennsylvania (this county
is the metropolitan area for Pittsburgh). Carlson conducted personal
interviews with each of the thirty-eight superintendents who headed
these school systems, asking each (1) in what year they had adopted
modern math, (2) which other superintendents were their best friends,
and (3) for certain other data. Modern math entered the local educa-
tional scene of Allegheny County by means of one school superinten-
dent, who adopted in 1958. This innovator traveled widely outside of
the Pittsburgh area, but he was a sociometric isolate in the local net-
work; none of the thirty-seven other school administrators talked with
him. The s-shaped diffusion curve did not take off until 1959-1960
after a clique of six superintendents adopted; these six included the
three main opinion leaders in the system. The rate of adoption then
began to climb rapidly. There was only one adopter in 1958 (the inno-
vator), five by the end of 1959, fifteen by 1960, twenty-seven by 1961,
A History of Diffusion Research 65
thirty-five by 1962, and all thirty-eight superintendents had adopted
by the end of 1963. Thus, modern math spread to 100 percent adop-
tion in about five years.
The cosmopolite innovator was too innovative to serve as an ap-
propriate role model for the other superintendents. They waited to
adopt until the opinion leaders in the six-member clique favored the
innovation.
Carlson's focus on interpersonal networks in diffusion repre-
sented a step forward from the Ryan and Gross (1943) hybrid corn
study, which failed to gather sociometric data. And the school super-
intendent study reminds one of the investigation of the diffusion of a
new drug among medical doctors, carried out by the medical sociol-
ogy tradition, which is discussed in the following section.
Public Health and Medical Sociology
This diffusion tradition began in the 1950s, about the same time that
medical sociology began to be recognized as a field of sociological spe-
cialization.* The innovations studied are (1) new drugs or other new
medical ideas, where the adopters are doctors, or (2) family-planning
methods or health innovations, where the adopters are clients or pa-
tients.
COLUMBIA UNIVERSITY DRUG STUDY
The classic study in this tradition was completed by three sociologists:
Elihu Katz, Herbert Menzel, and James Coleman, then of Columbia
University. This investigation is perhaps second only to the Ryan and
Gross analysis of hybrid corn in terms of its contribution to the diffu-
sion paradigm. The most noted impact of the Columbia drug study
was to orient future diffusion studies toward investigating the inter-
personal networks through which subjective evaluations of an innova-
tion are exchanged among individuals in a system. The drug study
helped illuminate the nature of diffusion networks, suggesting the role
Although many diffusion researchers in the public health and medical sociology
tradition do not necessarily identify themselves as "medical sociologists"; some are
affiliated with university schools of public health, for example, and others with
schools of medicine.
66 Diffusion of Innovations
that opinion leaders played in the "take off" of the s-shaped diffusion
curve.
The market research department of Charles Pfizer and Company,
a large pharmaceutical firm in New York City provided a grant of
about $40,000 to the Columbia sociologists for the project, which
began in 1954. A pilot study of the spread of a new drug was carried
out among thirty-three doctors in a New England town (Menzel and
Katz, 1955). The main investigation was conducted, after method-
ological techniques had been pretested in the pilot study, in four cities
in Illinois in late 1954.*
The drug study analyzed the diffusion of a new antibiotic that had
appeared in late 1953. The innovation was referred to by the Colum-
bia University researchers in most of their published reports by a
pseudonym, "gammanym." The drug had been tried at least once by
87 percent of the Illinois doctors, who had been using two other
closely related "miracle" drugs belonging to the same antibiotic fam-
ily as gammanym. The new drug superseded an existing idea just as
hybrid corn had replaced open-pollinated seed.
It is the patient rather than the doctor who pays for a new drug,
although it is the doctor who makes the innovation decision. The Col-
umbia University sociologists interviewed 125 general practitioners,
internists, and pediatricians in the four Illinois cities. These were 85
percent of the doctors practicing in specialities where 'the new drug
was of major potential significance" (Coleman et al, 1957). These 125
respondents sociometrically designated 103 additional doctors in
other specialities who were also interviewed. Whereas many of the
findings from the drug study are based upon the sample of 125 physi-
cians, the sociometric analyses of diffusion networks come from the
responses of the total sample of 228 doctors, which constituted 64 per-
cent of all doctors in active private practice in the four cities (Coleman
and others, 1957).
One of the neat methodological twists of the drug study was the
use of an objective measure of time of adoption from the written
record of drugstore prescriptions. The drug study is one of very few
diffusion investigations in which the researchers were not forced to
depend upon recall-type data on innovativeness. There was, in fact, a
The Illinois data are reported in Burt (1980), Menzel et al (1959), Coleman et al
(1957, 1959, and 1966), Katz (1956, 1957, and 1961), Katz et al (1963), and Menzel
(1957, 1959, and 1960). The present discussion features data mainly from the four Illi-
nois cities, rather than from the pilot study in New England.
A History of Diffusion Research 67
marked tendency for many doctors to report having adopted the drug
earlier than their prescription records indicated (Menzel, 1957),
although this might simply be because only a 10 percent sample of
prescription records was consulted by the diffusion scholars.
The Columbia University investigators were not aware of other re-
search traditions on diffusion at the time the gammanym data were
gathered. The researchers make no secret of their surprise upon dis-
covery of the hybrid seed study. Katz (1961) states: "The drug study
was completed ... without any real awareness of its many similarities
to the study that had been undertaken by Ryan and Gross almost fif-
teen years before."
Actually, there were some striking parallels between the hybrid
corn study and the drug study, given the considerable differences be-
tween farmers and physicians. For instance, innovative doctors at-
tended more out-of-town medical meetings than did later adopters,
reminding one of the innovative Iowa farmers who similarly displayed
their cosmopoliteness by visiting Des Moines (later diffusion studies
have also reported that innovators have friendship networks that ex-
tend outside of their local system). Just as the innovative Iowa farmers
had larger farms and higher incomes, the innovative doctors served
richer patients.
But the most important findings from the Columbia University
drug study, as already noted, dealt with interpersonal diffusion net-
works. Coleman et al (1966) found that almost all of the opinion
leaders, defined as the doctors who received three or more sociometric
choices as social friends, had adopted gammanym by the eighth
month (of the seventeen-month diffusion period that they studied). At
about this point, the s-shaped diffusion curve for the opinion leaders'
followers really took off. In other words, one reason for the s-shaped
curve is that once the opinion leaders in a system adopt, they then con-
vey their subjective evaluations of the innovation to their many net-
work partners, who are thereby influenced to consider adopting the
new idea.
Thus, a social system is a kind of collective-learning system in
which the experiences of the earlier adopters of an innovation, trans-
mitted through interpersonal networks, determine the rate of adop-
tion of their followers. Such "learning by doing" in a social system
could of course take a negative turn: if the new drug had not been very
effective in curing the innovative doctors' patients, they would have
passed their dissatisfactions with the new drug along to their peers.
 Diffusion of Innovations
68
Then the s-shaped diffusion curve would have displayed a much
slower rate of adoption. Or it might have reached a plateau and de-
clined as a result of widespread discontinuance.*
It is important to note that the doctors had plenty of information
about the new drug. Gammanym had undergone clinical trials by
pharmaceutical firms and by university medical schools prior to its
release to doctors. The results of these scientific evaluations of the in-
novation were communicated in medical journal articles to the physi-
cians in Coleman and others' (1966) sample, and by "detailmen"
(employees of the drug firms who contacted the doctors with informa-
tion about the new drug and who gave the doctors free samples of
gammanym). These communication messages created awareness
knowledge of the innovation among the medical community, but such
scientific evaluations of the new drug were not sufficient to persuade
the average doctor to adopt. Subjective evaluations of the new drug,
based on the personal experience by a doctor's peers, were key to con-
vincing the typical doctor to adopt its use with his own patients. When
an office partner said to his colleague: "Look, doctor, I prescribe
gammanym for my patients, and it cures them more effectively than
other antibiotics," that kind of message often had an effect.
This important research finding by Coleman and others (1966) led
the Columbia University sociologists to investigate which doctors
talked to whom. A doctor could talk to any one of the several hundred
other doctors in his community, but why did he choose the one, two,
or three other doctors as friends? A dyadic network analysis disclosed
that religion and age were the main determinants of friendship links,
with home town and the medical school attended also of some impor-
tance. But the main reasons for who-to-whom links in the medical
community were professional affiliations, such as belonging to the
same hospital or clinic as another doctor or else participating with him
or her in an office partnership. This finding suggested that the
organizational affiliations of medical doctors played an important
role in the diffusion of the medical innovation.
FAMILY PLANNING DIFFUSION IN DEVELOPING NATIONS
Since the classic investigation of drug diffusion, a considerable
number of other diffusion studies have been completed in the public
* The rate of adoption for the IUD contraceptive reached a plateau in India as a result
of negative rumors about the side effects of this innovation (Rogers, 1973, p. 300).
A History of Diffusion Research 69
health and medical sociology tradition.* Only a few of these studies
have dealt with the spread of new medical ideas to doctors; most are
investigations of the adoption of health or family planning innova-
tions by the public.
An important boost to the internationalization of the diffusion
field was the rise of "KAP surveys" in developing countries during
the 1960s. KAP studies are sample surveys of knowledge (K), at-
titudes (A), and practice (P) (that is, adoption) of family planning in-
novations. K, A, and P are the logical dependent variables in evalua-
tions of family planning communication campaigns, and as national
family planning programs arose after 1960 in many developing na-
tions (at first in Asia, then in Latin America, and finally in Africa) to
cope with the population problem, KAP-type diffusion researches
blossomed on all sides. Over 500 such KAP surveys were conducted in
72 nations by 1973 (Rogers, 1973, p. 377); India alone represented the
locale for over half of these investigations*.
With the exception of the Taichung experiment in Taiwan (Freed-
man and Takeshita, 1969), to be described shortly, the intellectual
contribution of these KAP surveys "to scientific understanding of
human behavior change has been dismal" (Rogers, 1973, p. 378). Al-
though they may not have advanced the diffusion model very much,
the KAP studies have served a useful function by generally showing
that most parents in developing countries want fewer children than
they actually have, and that the majority of the public desired a
government family planning program. Even the harshest critic of
KAP studies, Professor Philip H. Hauser (1967), stated that "KAP
survey results, erroneous or not, have helped to persuade prime minis-
ters, parliaments, and the general population to move in a desirable
direction and have provided family planning program administrators
with 'justification' for budgets and programs." So, the KAP surveys
had an important impact on policy makers in developing nations, ini-
tially showing that national family planning programs were feasible,
and later providing a means for evaluating the effectiveness of such
programs.
*The number of public health and medical sociology diffusion publications increased
from 36 in 1961 (7 percent of all diffusion publications), to 76 in 1968 (7 percent), to
226 in 1981 (7 percent); many of the recent studies in this tradition deal with family
planning methods in developing nations.
*Many of these fugitive studies could not be obtained and thus are not included in our
count of 226 publications in the public health and medical sociology tradition as of
1981.
 Diffusion of Innovations
70
Intellectually speaking, the family-planning diffusion studies were
generally disappointing,* although several modifications in the
classical diffusion model were formulated: the payment of incentives
to promote the diffusion and adoption of contraception, the use of
nonprofessional change agent aides, and the use of various communi-
cation strategies to help overcome the taboo nature of family plan-
ning. Such modifications in the classical diffusion model emerged
when family-planning programs in developing nations found the clas-
sical model wanting (Rogers, 1973).
THE TAICHUNG FIELD EXPERIMENT
Family planning diffusion studies gave a boost to field experiment *
research designs, as over a dozen such experiments in various nations
have followed the Taichung study in Taiwan (Rogers and Agarwala-
Rogers, 1975). This field experiment by Berelson and Freedman
(1964)* was one of the earliest and most important of the KAP
studies. Unlike the other KAP surveys, the Taichung study was a field
experiment, that is, an experiment conducted in the "real world"
rather than in the laboratory. In a field experiment, data are gathered
from a sample of respondents at two points in time by means of a
benchmark and a follow-up survey. Soon after the benchmark survey,
a treatment (or treatments) is applied to the sample. The effects of the
treatment can be determined by measuring the change in some vari-
able (for instance, adoption of innovations) between the benchmark
and the follow-up survey. One advantage of field experiment designs
is that they allow the researcher to determine the time order of his in-
dependent (treatment) variable on the dependent variable. As such,
field experiments are an ideal design for evaluating a diffusion pro-
gram. The Berelson and Freedman study in Taiwan was one of the
best, as well as one of the biggest: "This e f fo r t . . . is one of the most
* Particularly because of the unoriginal way in which the KAP studies were designed
and conducted. For example, the independent variables related to K, A, and P were
usually demographic variables like age, family size, formal education, and the like. It
was as if the researchers who conducted the KAP surveys did not realize that diffusion
is a particular type of communication process, in that mass media exposure and inter-
personal networks were not given much attention in the KAP surveys (Rogers, 1973,
pp. 379-389).
* A number of other publications report details on this research: Freedman (1964),
Freedman and others (1964), Freedman and Takeshita (1965, 1969), Gillespie (1965),
Takeshita (1964, 1966), and Takeshita et al (1964). All of these publications are sum-
marized in Freedman and Takeshita (1969).
A History of Diffusion Research 71
extensive and elaborate social science experiments ever carried out in a
natural setting" (Berelson and Freedman, 1964).
The researchers implemented four different communication inter-
ventions in approximately 2,400 neighborhoods (each composed of
twenty to thirty families) in Taichung, a city in Taiwan: (1) neighbor-
hood meetings about family planning, (2) neighborhood meetings,
plus mailed information about family planning to likely adopters, (3)
neighborhood meetings, plus a personal visit to the home of likely
adopters by a change agent who sought to persuade the women to
adopt family planning, and (4) neighborhood meetings, plus personal
visits by the change agent to both husband and wife in families likely
to adopt. In addition, all of the 2,400 neighborhoods in Taichung
were blanketed with family planning posters.
The results of this diffusion experiment were truly spectacular: 40
percent of the eligible audience of about 10,000 women adopted some
form of family planning. Pregnancy rates immediately decreased by
about 20 percent. Seventy-eight percent of the contraceptives adopted
were lUDs, the main family planning method promoted in the experi-
ment. The Taichung study showed that home visits by change agents
were essential for the success of a family-planning program. Mass
media communication (that is, the posters) created awareness-knowl-
edge, but interpersonal communication led more directly to adoption
of contraceptives. The Taichung researchers were surprised to find
that considerable interpersonal diffusion occurred between their
2,400 neighborhoods of study and the rest of the city (which was con-
sidered their control group). This unplanned diffusion spoiled their
neat experimental design, but it may have been their most important
finding. Again, we see that interpersonal networks among near-peers
energized the diffusion process.
The spectacular results of the Taiwan diffusion experiment pro-
vided optimism among development officials responsible for national
family planning programs which were then initiated in many develop-
ing countries. In the years since the Berelson-Freedman study, how-
ever, it was impossible to secure results comparable to those achieved
in Taiwan. So perhaps the Taiwan experiment led to an unrealistically
rosy glow about family planning diffusion, an optimism that was to be
dashed during the later 1960s and 1970s when many other nations
launched family-planning programs. In fact, the experience of these
programs to date suggests that contraceptives are one of the most dif-
ficult types of innovations to diffuse (Rogers, 1973) for reasons that
we will discuss in Chapter 6.
 Diffusion of Innovations
72
But the general point made by the Taichung family-planning study
is that diffusion researches need not be limited to conducting one-shot
surveys of the adopters of an innovation, with data gathering soon
after the new idea has diffused. A field experiment design allows a dif-
fusion researcher to draw on diffusion theory in order to plan one or
more communication interventions that can then be evaluated by ana-
lyzing data from benchmark and follow-up surveys. The results of dif-
fusion field experiments can lead both to advancing our understand-
ing of diffusion behavior, and to helping policy makers mount more
effective diffusion programs.
Communication
The communication tradition of diffusion research ranked as the sec-
ond largest (after the rural sociology tradition) by 1968, with eighty-
seven diffusion publications (8 percent of the total). At the time of my
1962 book, Diffusion of Innovations, there were only five diffusion
publications (1 percent of the total), and I did not even consider com-
munication as a diffusion research tradition. The rapid growth of the
communication tradition is shown by its position in 1981: 372 diffu-
sion publications, 12 percent of the total, and ranking second (again
to rural sociology) among the diffusion research traditions.
Diffusion research began before the academic field of communica-
tion research got underway. A focus on human communication as a
scientific field of study was not fully appreciated until an influential
book, The Mathematical Theory of Communication, was published
by Claude E. Shannon and Warren Weaver (1949); these scholars
identified the key concept of information and proposed a simple
model of communication. Then the field of communication research,
organized especially around studying the effects of mass communica-
tion, began to grow. At first, established scientists from political
science, sociology, social psychology, and other social science fields
were attracted to communication research. Soon, departments of
communication were established at many universities, and began pro-
ducing Ph.D.s in communication. These new scholars were often em-
ployed as professors in university departments of applied communi-
cation (such as journalism or advertising), and they helped inject an
emphasis upon communication theory and research into the existing
curricula.
Departments of journalism, speech, audio visual education, adver-
A History of Diffusion Research 73
tising, and television and radio broadcasting had, of course, existed
for many years. These departments mainly teach applied communica-
tion skills to undergraduates; they are primarily concerned with pro-
ducing professional communicators. Professors in these academic set-
tings realized that the training that they provide would be much more
useful and more academically respectable if it were based on scientific
research results and on communication theory. So they hired the new
Ph.D.s in communication research.
One of the early concerns of communication researchers was the
diffusion of news events carried by the mass media. Many such studies
have been completed, dealing with such headline news items as
Russia's launching of Sputnik, President Kennedy's assassination,*
and natural disasters. News events diffuse in a generally similar
fashion to technological innovations that have a material basis. The
distribution of knowers over time follows an s-shaped curve, interper-
sonal and mass-media channels play comparable roles, and so on. One
difference from the diffusion of other innovations is that news events
spread much more rapidly; for example, 68 percent of the U.S. adult
public was aware of the events in Dallas within thirty minutes of the
shot that felled the president. Soon thereafter, almost everyone knew
of this event.
In the early 1960s, communication researchers also began to inves-
tigate the transmission of technological ideas, especially agricultural,
health, educational, and family-planning innovations in developing
nations. Paul J. Deutschmann's study of the diffusion of innovations
in a Colombian village* stands as a landmark and led to a focus of
several communication researches upon peasant audiences in the
1960s. During the 1970s communication scholars began to investigate
the diffusion of technological innovations in the United States,
sometimes when communities or organizations were the adopting
units (Chapter 10).
One of the special advantages of the communication research tra-
dition is that it can analyze any particular type of innovation. There
are no limitations, such as the education tradition's focus on educa-
tional innovations, the rural sociologist's main emphasis upon agri-
* Greenberg's (1964) analysis of the diffusion of the news of the Dallas assassination is
somewhat typical of the approach used in the news event diffusion studies. Probably
the most noted news event diffusion study, however, is by Deutschmann and
Danielson (1960); it set the pattern for the other news diffusion studies that followed.
*The publications from this study are Deutschmann (1963), Deutschmann and Fals
Borda (1962a,b), and Deutschmann and Havens (1965).
74 Diffusion of Innovations
cultural ideas, or the medical sociologist's concern with family-plan-
ning methods. This lack of a message-content orientation perhaps
allows the communication researcher to concentrate on the process of
diffusion. Further, the communication tradition has an appropriate
toolkit of useful concepts and methods (for example, credibility, net-
work analysis, and the semantic differential) for studying diffusion.
In fact, the enthusiastic way in which communication scientists have
taken to diffusion research makes one wonder why they did not do so
even sooner. The multidisciplinary backgrounds of communication
research help this tradition integrate the work of various other diffu-
sion research traditions.
Marketing
Another diffusion tradition that came on strong in the 1960s and
especially in the 1970s, is marketing. Marketing managers of firms in
the United States have long been concerned with how to launch new
products most efficiently. Their interest in this topic is sparked by the
regular launching of large numbers of new consumer products, many
of which fail. For instance, it is estimated that only one idea out of
every 540 results in a successful new product (Marting, 1964, p. 9).
Only 8 percent of the approximately 6,000 new consumer items intro-
duced each year have a life expectancy of even one year (Conner,
1964). Commercial companies, therefore, have a vital stake in the dif-
fusion of new products, and a great number of such researches have
undoubtedly been completed. A large proportion of these diffusion
research reports, however, are found only in the secret files of the
sponsoring companies. Unfortunately, the funding of marketing dif-
fusion studies by private sources, who wish to use the results to gain a
competitive advantage, leads to restrictions on scholarly access to the
intellectual lessons learned from these studies in the marketing tradi-
tion.
Even so, the available research literature in the marketing tradition
is quite impressive today. In 1961, there were only a handful of
marketing diffusion studies and I did not then consider that a market-
ing tradition existed (Rogers, 1962). By 1968, however, I could iden-
tify sixty-four marketing diffusion publications, 5.9 percent of the
total (Rogers with Shoemaker, 1971). By 1981, there were 304 market-
ing publications, 10 percent of the total, and marketing ranked fourth
in its contribution to diffusion research (Table 2-2)! Our tabulation is
A History of Diffusion Research 75
complemented by Glazer and Montgomery (1980) who carried out a
comprehensive literature search of twenty-five relevant journals in
economics, marketing, and general management from 1960 to 1979.
They found 407 articles and 81 books, a total of 488 publications, that
dealt with the diffusion of innovations.* This literature emphasized
the test marketing of new products, identifying markets for new prod-
ucts, strategic planning for diffusion, and studies of how the per-
ceived attributes of an innovation affect its purchase. Clearly, the dif-
fusion approach has caught on in the field of marketing.
"Marketing" has a pejorative ring in some academic circles be-
cause the term is narrowly construed as synonymous with manipulat-
ing human purchasing behavior for commercial advantage (Rogers
and Leonard-Barton, 1978). Marketing scholars do not deny that
some marketing efforts are conducted to try to sell products to people
who do not really want them. But they argue that most marketing ac-
tivities, if they are to be very successful, must match consumers' needs
and desires with commercial products and services. In fact, marketing
researchers argue that they are providing a useful contribution to
society by helping to identify consumer needs, and by fulfilling such
needs by making commercial products available.
The marketing approach can also be applied to selling noncom-
mercial products, in what is called "social marketing" (Kotler and
Zaltman, 1971). Here the objective is to diffuse socially beneficial
ideas that do not entail the sale of commercial products. Social mar-
keting was launched about thirty years ago with the rhetorical ques-
tion "Why can't you sell brotherhood like you sell soap?" (Wiebe,
1952). In the past decade or so, the social marketing approach has
been applied to such causes as energy conservation, antismoking,
safer driving, family planning, preventing drug abuse, and improving
nutrition. Often social marketing campaigns seek to convince people
to do something that is unpleasant. For instance, a recent survey
showed that nine out of ten smokers in the United States said they
would like to quit, yet 57 percent expected that they would still be
smoking in five years' time. Likewise, many individuals wish to lose
weight, get more exercise, and floss their teeth, but they do not. The
mam applications of social marketing, then, are to changing behav-
iors in directions desired by the individuals involved, but that seem to
be impeded by inertia.
* The Glazer and Montgomery (1980) tabulation differs somewhat from ours because
of their use of somewhat different criteria for diffusion studies.
76 Diffusion of Innovations
An assessment of the past decade of experiences with social mar-
keting by Fox and Kotler (1980) concludes "that most social market-
ing problems will be more formidable than the typical marketing
problems facing commercial marketers." One of the greatest suc-
cesses for social marketing has been its use by government family-
planning programs to diffuse birth control pills and condoms in India,
Kenya, Sri Lanka, Thailand, Bangladesh, and Mexico (Rogers, 1973).
For example, the condom campaign in India in the early 1970s in-
volved renaming the product as "Nirodh" (from a Sanskritic word
meaning "protection"); it had been known as "French letter" or
"FL," rather taboo terms. After trying out their Nirodh campaign in
a small test market near New Delhi, the social marketers expanded
their coverage to one-fifth of India, and then by careful stages to the
entire nation. A massive advertising campaign helped launch Nirodh,
and the condoms were sold by thousands of teashops and at cigarette
stands on every street corner. The government of India subsidized the
product so that each condom only cost about two cents. Market re-
search was conducted at every step of the Nirodh campaign to provide
feedback for decisions by the campaign planners: the selection of the
name Nirodh over various alternatives, which kind of distribution
outlet would be most accessible and most acceptable to the intended
audience, and what information was needed by Indian men about how
to use condoms. Thus, the Nirodh campaign shows how marketing ex-
pertise, along with diffusion strategies, were used in this social
marketing activity.
The marketing tradition of diffusion research has certain advan-
tages and some attendant disadvantages compared with other research
traditions. Because marketing scholars usually conduct diffusion
studies with the sponsorship, or at least the collaboration, of the
manufacturers of a new product, the researchers are able to conduct
field experiments (an especially powerful type of diffusion research
design, as we discussed previously). Other than in marketing, diffu-
sion scholars have seldom been in a position to control the interven-
tion strategies through which an innovation is introduced, so it has not
been possible to conduct field experiments. In fact, several field ex-
periments on diffusion have been conducted by marketing scholars
(for example, Arndt, 1967b, 1971).
But such close siding with the sources of innovations in diffusion
research can also bring with it some intellectual and ethical problems.
For example, the diffusion problems and needs of marketers are usu-
ally given priority over those of the consumers. Sources often wish to
A History of Diffusion Research 77
know how they can influence the consumers' adoption behavior. In
contrast, consumers may wish to know how to insulate themselves
from such influence attempts or, more generally, how they can evalu-
ate new products (Rogers and Leonard-Barton, 1978). The source bias
in many marketing diffusion studies may lead to highly applied
research that, although methodologically sophisticated, deals with
trivial diffusion problems. As a result, we may know more about con-
sumer preferences for deodorant scents and the taste of beer than
about product safety, or about how best to advance the theory of dif-
fusion.
Geography
Although still one of the smallest of the nine main diffusion research
traditions described in this book, the geography tradition has ex-
panded considerably in recent years, and it is unique in its emphasis
upon space as a factor affecting the diffusion of innovations.
In 1961, there were only three diffusion publications in geography,
all by Dr. Torsten Hagerstrand at the University of Lund in Sweden
(Rogers, 1962). By 1968, there were only seven publications in this
tradition (0.6 percent of the total), with the four new studies being
conducted in the U.S. (Rogers with Shoemaker, 1971). In 1981, there
were 130 diffusion publications by geographers, representing about 4
percent of the total.
One's stereotype of the field of geography probably recognizes
that maps are one of the geographers' favorite tools. Space is the
crucial variable for geographers, and they specialize in investigating
how spatial distance affects all other aspects of human existence. Pro-
fessor Hagerstrand (1952, 1953) pioneered a simulation approach to
investigating how spatial distance affected diffusion. First, Hager-
strand constructed a mathematical model of the diffusion process as it
should theoretically occur over time and through space. For instance,
Hagerstrand's model contained, as one of its elements, the "neigh-
borhood effect," which expressed the tendency for an innovation to
be more likely to spread from an adopter to another adopter (in the
next unit of time) who was close by, rather than far away. This neigh-
borhood effect was built into Hagerstrand's computer model of diffu-
sion by means of mathematical probabilities (of adoption) that de-
creased with distance away from the adopter. Hagerstrand then
entered a map of the Swedish countryside in his computer, and, begin-
78 Diffusion of Innovations
ning with the location of the first adopter of an agricultural innova-
tion, he simulated the diffusion process. He then compared the result-
ing simulation of diffusion with data on the actual rate of adoption
and geographical spread of the farm innovation.
The basic research approach of diffusion simulation is an attempt
to mimick the reality of diffusion. If the simulated process does not
correspond to the reality data, then the researcher must adjust his
theoretical model of diffusion and try again. Hagerstrand remained
the dean of the diffusion simulation approach, and for more than a
decade few other researchers took up his novel and interesting ap-
proach. Not until the mid-1960s did a set of quantitative geographers
in the United States begin to pick up on the simulation approach and
carry it forward in a series of research studies.
American geographers also began to pursue nonsimulation diffu-
sion research, but always with emphasis on the spatial variable. The
results show clearly that space is important in determining the adop-
tion of an innovation. Dr. Lawrence A. Brown of Ohio State Univer-
sity has become a leading figure in the geography diffusion tradition
during the 1970s, publishing a number of research papers and an im-
portant book (Brown, 1981).
General Sociology
The general sociology tradition of diffusion research is a somewhat
residual category, consisting of all other diffusion studies not included
in early sociology, rural sociology, and medical sociology. In my
previous two books on diffusion, the number of diffusion publica-
tions by general sociologists did not justify their consideration as a
major diffusion tradition. But since the late 1960s, diffusion studies
by general sociologists have proliferated; in 1981, this research tradi-
tion included 282 diffusion publications, 9 percent of the total (see
Table 2-2). General sociology had climbed to fifth place among the
diffusion research traditions.
The rise of general sociology as a research tradition indicates that
the diffusion approach is catching on among many sociologists today,
not just those concerned with agricultural or medical or health inno-
vations.
Table 2-2 shows six of the minor research traditions on diffusion,
in addition to the nine traditions just discussed: general economics,
political science, agricultural economics, psychology, statistics, and
industrial engineering. These and other minor traditions make up 500
A History of Diffusion Research 79
of the 3,085 diffusion publications available in 1981, or about 17 per-
cent. Today all of the behavioral science disciplines are represented by
at least a certain degree of interest in the diffusion of innovations.
We can expect further minor diffusion traditions to develop in the
future as the diffusion approach continues to spread to other disci-
plines. But at the same time we expect a more complete merger of the
existing traditions, at least at the conceptual and methodological level.
A Typology of Diffusion Research
When showing a large city to a stranger it is often wise to take the
visitor first to the top of a skyscraper so that he may scan the entire
landscape prior to being immersed in the details of the city. Likewise,
in this section we hope to provide the reader with an overall impres-
sion of types of diffusion research before we move to a more detailed
discussion in later chapters. Our present concern differs from the
previous discussion of the history of diffusion research in that we now
shall look at types of diffusion research, rather than at the traditions.
Table 2-3 shows eight different types of diffusion analysis that
have been completed and the relative amount of attention paid to
each. By far the most popular diffusion research topic has been vari-
ables related to individual innovativeness (type 3 in Table 2-3). More
than half (58 percent) of all the empirical generalizations reported in
available diffusion publications deal with innovativeness. We il-
lustrate each of these eight types of diffusion research with one or two
studies, in order to convey the nature of such diffusion investigations.
1. Earliness of knowing about innovations. Greenberg (1964)
determined what, when, and how people first learned about the news
of the assassination of President Kennedy. Data were gathered by
telephone interviews with 419 adults in a California city. The respon-
dents were classified as "early knowers" or "late knowers." Most of
the early knowers reported that they had heard of Kennedy's death by
radio or television, whereas most of the late knowers first learned of
the assassination by means of interpersonal communication channels.
Most of the individuals who first learned about this news event from a
mass medium then told other individuals about the message. And
most individuals who first learned about the news through an in-
terpersonal network then turned to a mass media channel for further
information and to obtain confirmation of the news event.
2. Rate of adoption of different innovations in a social system.
Table 2-3. Types of Diffusion Research.

* These percentages are based on a content analysis of the 6,811 generalizations identified in the diffusion literature available in 1968, which consisted of 1,084
publications reporting empirical research results. As of 1981 there were 3,085 empirical publications available, but our impression (based on reading these
studies) is that the percentages reported in this table have not changed much since 1968 (when the complete content analysis was conducted).
 Diffusion of Innovations
82
Fliegel and Kivlin (1966b) conducted personal interviews with 229
Pennsylvania dairy farmers. The investigation used farmers' percep-
tions of fifteen attributes of each of thirty-three dairy innovations to
predict the rate of adoption for this sample of Pennsylvania farmers.
Innovations perceived as most economically rewarding and least risky
were adopted more rapidly. The complexity, observability, and trial-
ability of the innovations were less highly related to the rate of adop-
tion, but innovations that were more compatible with farmers' values
were adopted more rapidly.
3. Innovativeness. Deutschmann and Fals Borda (1962b) con-
ducted a diffusion survey in a Colombian village to test the cross-
cultural validity of correlates of innovativeness derived from prior
U.S. diffusion research. The primary hypothesis of the study was that
after taking cultural differences into account, the basic pattern of dif-
fusion of new farm ideas would be substantially the same in Saucio
(the Colombian village) as in the United States. A striking similarity
was found between the results obtained from the Colombia study and
those reported for Ohio farmers (by Rogers, 1961): The characteristics
of innovators such as greater cosmopoliteness, higher education, and
larger-sized farms were remarkably similar in Saucio and in Ohio.
Another variant of the correlates-of-innovativeness study is
Mohr's (1969) survey of the directors of county departments of public
health in Michigan, Ohio, and Ontario (Canada). An innovativeness
score was computed for each of the 120 health departments of study,
indicating the degree to which each organization had adopted various
new ideas in public health. The most innovative health departments
were characterized by more financial resources, a director who was
more highly committed to innovation, and larger size.
4. Opinion leadership. The success or failure of diffusion pro-
grams rests in part on the role of opinion leaders and their relationship
with change agents. Rogers and van Es (1964) sought (1) to identify
opinion leaders in five Colombian villages; (2) to determine their
social characteristics, communication behavior, and cosmopoliteness;
and (3) to determine the differences in these correlates of opinion
leadership on the basis of systems with different norms. The data were
gathered in personal interviews with 160 peasants in 3 modern villages
and with 95 peasants in two traditional communities. Rogers and van
Es found that opinion leaders, when compared to their followers in
both modern and traditional systems, were characterized by more for-
mal education, higher levels of literacy, larger farms, greater innova-
tiveness, higher social status, and more mass-media exposure. In the
A History of Diffusion Research 83
modern villages, however, the opinion leaders were young and inno-
vative, reflecting the norms, whereas in the traditional systems the
leaders were older and not very active in adopting new ideas (Rogers
with Svenning, 1969). Thus, the leaders tended to reflect the norms of
their village.
5. Who interacts with whom. Rogers and Kincaid (1981, pp. 303-
SOS) conducted personal interviews with the sixty-nine married
women in a Korean village in order to determine the role of interper-
sonal networks in the diffusion of family-planning innovations. Each
respondent was asked which other women she talked with about con-
traceptive methods. Spatial location of each respondent's home was a
very important predictor of who talked with whom, even though the
village was extremely small (only about two typical city blocks in
diameter). But space was by no means a complete explanation of dif-
fusion networks links; in fact some women talked with a peer on the
opposite side of the village. Physically lengthy links were especially
characteristic of opinion leaders, which suggested that one of the im-
portant roles of such leaders was to interconnect the spatially related
cliques in the village, and thus to increase the connectedness of the
village's communication structure. Social similarity also helped ex-
plain who was linked to whom; women of similar social status and age
were more likely to interact with each other. A general conclusion
from who-to-whom studies is that space and social distance (that is,
heterophily/homophily) are the main determinants of who talks to
whom in diffusion networks.
We have already mentioned in this chapter the drug diffusion
study by Coleman et al (1966). As one part of their investigation, these
diffusion scholars asked their respondents to name the other doctors
who were their best friends. Coleman et al then determined the main
variables that explained who talked to whom in network links. Simi-
larity in age, religion, hometown, and the medical school attended
were important factors structuring who talked to whom. But the most
important variables determining who-to-whom links in the medical
community were such professional affiliations as practicing in the
same clinic, hospital, or office partnership. Doctors were more likely
to talk about the new drug if they worked together.
6. Rate of adoption in different social systems. Rogers and Kin-
caid (1981, pp. 279-281) sought to explain the rate of adoption of
family planning innovations in twenty-four Korean villages. Unlike
diffusion research type 2, where the purpose is to explain why some in-
novations have a faster rate of adoption than others, in this type of
 Diffusion of Innovations
84
research we study why the same innovation is adopted more rapidly in
certain systems than it is in others. The Korean villages with the fastest
rates of family-planning adoption were composed of families with
higher mass-media exposure to family planning, had leaders with
more highly connected networks in their village, and were villages
with more change agent contact. The economic resources of the
village were less important in explaining rate of adoption.
7. Communication channel usage. The Ryan and Gross (1943) in-
vestigation of the diffusion of hybrid-seed corn in Iowa found that the
typical Iowa farmer first heard of hybrid seed from a commercial
salesman but that neighbors were the most influential channel in per-
suading a farmer to adopt the innovation (although later research has
generally shown that salesmen are not the most important channel at
the knowledge stage). Ryan and Gross were the first researchers to
suggest that an individual passes through different stages (knowledge
and persuasion, for example) in adopting a new idea. Different com-
munication channels play different roles at these various stages in the
innovation-decision process. Salesmen were more important channels
about the innovation for earlier adopters, and neighbors were more
important for later adopters. This finding suggests that communica-
tion channel behavior is different for the various adopter categories, a
proposition that is supported by later diffusion researches.
8. Consequences of innovation. The consequences of the use of
the steel ax by a tribe of aborigines were studied by Sharp (1952). The
Yir Yoront were relatively unaffected by modern civilization, owing
to their isolation in the Australian bush, until some missionaries
moved in nearby. They distributed steel axes among the Yir Yoront as
gifts and as pay for work performed. Before the introduction of the
steel ax, the stone ax had served as the Yir Yoront's principal tool and
as a symbol of masculinity and respect. Only men could own stone
axes, so the women and children, who were the main users of these
tools, borrowed them according to a system prescribed by custom.
But the missionaries gave axes to anyone. The steel axes caused a ma-
jor disruption of Yir Yoront culture, and a revolutionary confusion of
age and sex roles. Elders, once highly respected, now became depen-
dent upon women and younger men for steel axes. The consequences
of the steel ax were unanticipated, far-reaching, and disruptive (as we
shall detail in Chapter 11).
The reader has now been provided with a brief glimpse of the dif-
fusion landscape in terms of the eight directions in which it has been
growing. In later chapters of this book, we shall probe these eight
A History of Diffusion Research 85
types of diffusion research in much greater detail. We hope that the
typology of diffusion research just discussed, although brief, will pro-
vide the reader with an overall research map of the entire field.
Summary
A theme of the present chapter is that although diffusion research
began as a series of scientific enclaves, it has emerged in recent years as
a single, integrated body of concepts and generalizations, even though
the investigations are conducted by researchers in several scientific
disciplines. A research tradition is a series of investigations on a
similar topic in which successive studies are influenced by preceding
inquiries. Nine major diffusion traditions are described: anthropol-
ogy, early sociology, rural sociology, education, medical sociology,
communication, marketing, geography, and general sociology.
Eight main types of diffusion research are identified, and dealt
with in detail in future chapters:
1. Earliness of knowing about innovations.
2. Rate of adoption of different innovations in a social system.
3. Innovativeness.
4. Opinion leadership.
5. Who interacts with whom in diffusion networks.
6. Rate of adoption in different social systems.
7. Communication channel usage.
8. Consequences of innovation.
Our tour in this chapter of the past forty years of diffusion
research provides many examples of Thorsten Veblen's concept of
"trained incapacity": by being taught to "see" innovativeness, opin-
ion leadership, and other aspects of the classical model of diffusion,
we failed to "see" much else. Acceptance of an intellectual paradigm
by scholars in a research field enables them to cope with uncertainty
and information overload, through the simplification of reality that
the paradigm represents. It also imposes and standardizes a set of
assumptions and conceptual biases that, once begun, are difficult to
recognize and overcome. That is the challenge for the next generation
of diffusion scholars.
A critical statement that appeared in my first book on diffusion
(Rogers 1962, p. x) is perhaps still fitting today, twenty years later: